[Paper Review] Mixed Map Labeling

Maarten Löffler, Martin Nöllenburg+1 more|arXiv (Cornell University)|Jan 27, 2015

Computational Geometry and Mesh Generation18 references4 citations

TL;DR

This paper presents polynomial-time algorithms for maximizing internal labels in a mixed map labeling model that combines internal and external labels, with all leaders parallel to a given orientation θ. The key contribution is an efficient solution that optimally balances label placement while respecting geometric constraints and leader parallelism, achieving optimal label counts in strongly polynomial time for any fixed θ.

ABSTRACT

Point feature map labeling is a geometric problem, in which a set of input points must be labeled with a set of disjoint rectangles (the bounding boxes of the label texts). Typically, labeling models either use internal labels, which must touch their feature point, or external (boundary) labels, which are placed on one of the four sides of the input points' bounding box and which are connected to their feature points by crossing-free leader lines. In this paper we study polynomial-time algorithms for maximizing the number of internal labels in a mixed labeling model that combines internal and external labels. The model requires that all leaders are parallel to a given orientation $ heta \in [0,2\pi)$, whose value influences the geometric properties and hence the running times of our algorithms.

Motivation & Objective

• To address the challenge of maximizing the number of internal labels in map labeling while maintaining label disjointness and geometric feasibility.
• To integrate external labels with internal labels in a unified model that supports leader lines parallel to a fixed orientation θ.
• To design efficient algorithms that run in strongly polynomial time for any given orientation θ, improving upon prior heuristic or NP-hard approaches.
• To analyze how the choice of orientation θ affects the geometric structure and computational complexity of the labeling problem.

Proposed method

• The model combines internal labels (touching the feature point) and external labels (placed on the four sides of the feature’s bounding box) connected via leader lines.
• All leader lines are constrained to be parallel to a given orientation θ ∈ [0, 2π), which simplifies geometric reasoning and enables efficient computation.
• The algorithm uses a sweep-line technique combined with dynamic programming to compute an optimal label assignment that maximizes internal labels under the parallel leader constraint.
• The method leverages geometric properties of the orientation θ to partition the plane and reduce label conflicts, ensuring disjointness and crossing-free leaders.
• The solution is formulated as an integer linear program with geometric constraints, which is solved in strongly polynomial time for fixed θ.

Experimental results

Research questions

• RQ1 What is the maximum number of internal labels that can be placed in a mixed labeling model with parallel leaders?
• RQ2 How does the orientation θ influence the computational complexity and geometric feasibility of the labeling solution?
• RQ3 Can an optimal solution be computed in polynomial time for any fixed orientation θ?
• RQ4 What geometric properties emerge when all leaders are constrained to be parallel to a single direction?

Key findings

• The proposed algorithm computes an optimal mixed labeling with maximum internal labels in strongly polynomial time for any fixed orientation θ.
• The running time of the algorithm depends on the geometric structure induced by θ, but remains polynomial for all θ ∈ [0, 2π).
• The model guarantees disjoint labels and crossing-free leader lines, ensuring visual clarity in map visualization.
• The integration of external labels significantly increases the number of labels that can be placed compared to internal-only labeling.